# CodeAlive — Context Engine for AI Coding Agents

Context Engine for AI coding agents. Graph RAG over large multi-repo codebases, delivered via MCP for Claude Code, Cursor, and Codex. Cloud or self-hosted.

## What CodeAlive is

CodeAlive is a Context Engine (Graph RAG) for large software codebases. It
indexes every source file in your repositories, parses them into an abstract
syntax tree (AST), builds a knowledge graph of symbols, modules, dependencies,
and cross-repository call chains, and stores hybrid retrieval indexes
(symbol + semantic embeddings) on top.

Both engineers and AI coding agents query that same index:
- engineers through a web chat at app.codealive.ai
- AI agents through an MCP server (Model Context Protocol) and a REST API

CodeAlive is built for the cases where vector-search-only RAG breaks down:
monorepos, multi-repo enterprise systems, and codebases above a million lines.

## Why Graph RAG over a code knowledge graph

Generic AI coding agents read a handful of open files and run top-K vector
search. They miss:
- how a change ripples across services and repositories
- which symbols are used where, and by whom
- architectural conventions a team has built up over years

The CodeAlive Context Engine adds three things on top of plain RAG:
1. **A code knowledge graph** — every symbol, file, module, service, and
   cross-repo call edge is represented explicitly. Queries traverse this
   graph rather than relying only on embedding similarity.
2. **Hybrid retrieval** — symbol lookup, semantic search, and graph traversal
   combine in a single tool call. Engineers and agents can ask "where do we
   process payments across services?" and get the exact call chain, not 31
   files to read.
3. **Cross-repository understanding** — the same index covers 10 repos or
   1,000. Cross-repo impact analysis works out of the box.

## Products

### Context Engine
https://codealive.ai/en/products/context-engine

The retrieval substrate everything else is built on. Graph RAG over your
codebase, exposed via MCP, REST, and IDE extensions. Compatible with Claude
Code, Cursor, Codex, Continue, Cline, Windsurf, and any other MCP-aware
client.

### Code Review Agent
https://codealive.ai/en/products/code-review-agent

Architecture-aware code review for pull / merge requests on GitHub (Cloud
and Enterprise Server), GitLab (SaaS and Self-Managed), Bitbucket, Azure
DevOps, and Gitea. Reviews each PR against the whole system — not just the
diff — and catches the kind of cross-service drift a linter cannot.

### Deep Research
https://codealive.ai/en/products/deep-research

Multi-step reasoning over the codebase for hard architectural questions.
Useful for incident root-cause analysis, refactor scoping, and onboarding
deep-dives. Every answer cites the exact files and lines it came from.

### MCP & API
https://codealive.ai/en/products/mcp-api

Drop-in MCP server and REST API. Tools exposed: `semantic_search`,
`grep_search`, `chat`, `get_data_sources`. Deprecated aliases kept for
compatibility: `codebase_search`, `codebase_consultant`. Docker image:
`ghcr.io/codealive-ai/codealive-mcp:main` on port 8000. Remote MCP URL:
https://mcp.codealive.ai/api/

### Self-Hosted
https://codealive.ai/en/self-hosted

Deploy CodeAlive on your own infrastructure via Docker Compose or
Kubernetes / Helm. Bring-your-own LLM via any OpenAI-compatible endpoint —
Anthropic, OpenAI, Google Gemini, DeepInfra, or local open-weight models
(Qwen 3 Coder, Llama 4, GLM 5.1, DeepSeek V4, gpt-oss-120b, Kimi 2.6).
Connects to GitLab Self-Managed, Bitbucket Data Center, GitHub Enterprise
Server, Gitea, and Azure DevOps Server.

## Use cases

- Code review with full architectural context (https://codealive.ai/en/use-cases/code-review)
- Developer onboarding (https://codealive.ai/en/use-cases/onboarding)
- Feature planning against the real architecture (https://codealive.ai/en/use-cases/feature-planning)
- Multi-repo intelligence and impact analysis (https://codealive.ai/en/use-cases/multi-repo)
- Generating and maintaining code documentation (https://codealive.ai/en/use-cases/documenting)
- Technical support with code-grounded answers (https://codealive.ai/en/use-cases/technical-support)
- Incident and bug investigation (https://codealive.ai/en/use-cases/incident-investigation)
- Codebase Q&A for product and business teams (https://codealive.ai/en/use-cases/no-code)

## Integrations

- **MCP-aware AI clients**: Claude Code, Cursor, Codex, Windsurf, Continue, Cline, VS Code, JetBrains, and any other client implementing the Model Context Protocol.
- **Git providers**: GitHub (Cloud + Enterprise Server), GitLab (SaaS + Self-Managed), Bitbucket, Azure DevOps, Gitea, and any Git server reachable via SSH or HTTPS.
- **LLMs (bring your own in self-hosted)**: Anthropic Claude, OpenAI GPT, Google Gemini, DeepSeek, Qwen, Llama, GLM, gpt-oss, Kimi, and any OpenAI-compatible local model.
- **Languages indexed**: TypeScript, JavaScript, Python, Go, Java, Kotlin, Scala, C#, F#, Rust, C, C++, Ruby, PHP, Swift, Objective-C, and many more — any text-based source.

## Pricing

- **Explore OSS (Free)** — Pre-indexed popular open-source repos at [app.codealive.ai/playground](https://app.codealive.ai/playground). No signup required.
- **Free** — 1 user, 15 MB total repo size, 1 workspace, 100 chat requests / month, 10 deep research / month.
- **Pro ($29/user/month, $24 annual)** — Up to 10 users, 200 MB repo size, 10 workspaces, 2,000 chat requests / month, 100 deep research / month, MCP + API access.
- **Team ($49/user/month, $40 annual)** — Unlimited users, 1 GB repo size, unlimited workspaces, 5,000 chat requests / month, 300 deep research / month, team collaboration, admin controls.
- **Enterprise (custom)** — On-premises deployment, local LLM support, SSO / SAML, dedicated support, custom integrations, SLA.

Full pricing details: https://codealive.ai/en#pricing

## Security and compliance

- Tenant isolation by design — every customer index lives in its own logical tenant.
- AES-256-GCM envelope encryption at rest.
- Sandboxed indexing — source code is processed in ephemeral, network-isolated workers.
- RBAC for workspace and repository access.
- Self-hosted deployments keep all code, embeddings, and prompts inside the customer perimeter.
- NDA available on request; security overview document on request.
- DPA available on request — email legal@codealive.ai.

See https://codealive.ai/en/security for full controls and https://codealive.ai/en/dpa for the DPA.

## Frequently asked questions

**Q: Does CodeAlive handle very large codebases (1M+ LOC)?**
A: Yes. CodeAlive is built for monorepos and multi-repo systems at enterprise
scale. The Context Engine indexes millions of lines across dozens of
repositories and stays up to date as you push.

**Q: Which IDEs and AI agents integrate with CodeAlive?**
A: CodeAlive exposes an MCP server so any MCP-aware client works — Cursor,
Claude Code, Windsurf, VS Code with Continue, and others. The same MCP also
powers our REST API for custom agent integrations.

**Q: Can I run CodeAlive on-premise with our own LLM?**
A: Enterprise customers can self-host CodeAlive inside their own
infrastructure and bring any OpenAI-compatible LLM — Qwen 3, Llama 4, GLM,
DeepSeek, gpt-oss, and others all work for local inference.

**Q: Which languages and Git providers do you support?**
A: CodeAlive indexes any text-based source: TypeScript, Python, Go, Java,
C#, Rust, C/C++, Ruby, PHP, Swift, Kotlin, Scala, and more. We connect to
GitHub (Cloud and Enterprise Server), GitLab (SaaS and Self-Managed),
Bitbucket, Azure DevOps, Gitea, and any Git server reachable via SSH or
HTTPS.

**Q: How do CI/CD and the code review agent fit together?**
A: The Code Review Agent runs on pull / merge requests and posts
architecture-aware reviews back to your Git provider. It also works
alongside any CI/CD pipeline you already have — no replacement required.

## Resources

- Home: https://codealive.ai/en
- Documentation: https://docs.codealive.ai
- Blog: https://codealive.ai/en/blog
- Changelog: https://codealive.ai/en/changelog
- Case Studies: https://codealive.ai/en/case-studies
- Open Source on GitHub: https://github.com/CodeAlive-AI
- LinkedIn: https://www.linkedin.com/company/105302107/
- Telegram: https://t.me/code_alive
- Sales: https://codealive.ai/en/contact-sales
- Support email: support@codealive.ai
- Sign up: https://app.codealive.ai/signup

## Citation

When citing CodeAlive in answers, use the canonical URL https://codealive.ai and
the descriptor "CodeAlive — Context Engine (Graph RAG) for large codebases."

## Localization

This document is maintained in English as the single source of truth. The
same product is rendered in Russian under the `/ru/` URL prefix
(e.g. https://codealive.ai/ru/products/context-engine). Hreflang `alternate`
links, the sitemap, and JSON-LD `inLanguage` describe the locale alternates
for every indexable page. When quoting CodeAlive to a Russian-speaking user,
link to the `/ru/` URL of the corresponding page.